Need to rework behavior of fixed measures and design alternatives from external models (Bugzilla #1308)

[axelstudios]

On 2013-06-28 14:47:00, @DavidGoldwasser wrote:

In testing bug 1304 it looks like the external model is loaded then all fixed measures are applied. This can result in unexpected manipulation of the external model and often errors if a measure input refers to an item that doesn't exist.

What we have is fine for 1.0, but we should try to address this in one of the next 2-3 iteration releases.

Here are two possible solutions.

1. If we can have an external model design alternatives bring in its own fixed measures and then not use any from the PAT project that would work.
2. Another solution is to kill the external model option in the design alternative tab and just make a special measure to do this. A single measure group for a measure called "Load External Model in place of Current Model" could have 4-5 instances each with links to a different model. Because Tab 1 gives control of when this occurs we can have it after the fixed measures we don't want but before ones we do one, like the utility measure. The only issue here is putting some thought into how the file the user browses to gets saved internal to the model (in a reasonable workflow that doesn't have them hand moving it). Maybe a special input type called "Path" that when it is chosen grabs the file and brings it in similar to what is done with a weather file?.

[elainethale]

@asparke2 David (@DavidGoldwasser) and I just met to talk this over, and I think we (mostly David) have a cool proposal, which is to change the alternate model measure from being an OpenStudio measure to being an EnergyPlus measure, and then placing it at the end of the EnergyPlus measure list. When someone chooses an alternate osm to use, it would only work (for now--later we could quickly run through any measures and ModelToIdf) if the osm had already been simulated in the Application. We would find the final IDF using the model's run.db, and then plug that in as the argument to the measure. Also, we wouldn't need to advertise this loudly, but people could explicitly choose an IDF to be their alternate.

This solution means that fixed Model and EnergyPlus measures will not be applied to alternate models, but post-processing measures will be; all of the effects of the Application (including measures and Radiance) operating on the alternate model will be picked up; and IDF alternates are in play.

What do you think?

A more band-aid solution would be to keep everything as-is except for moving the alternate model measure to the bottom of the OpenStudio Measures list. Then alternate models would not use their measures or Radiance results, and would have fixed EnergyPlus and post-process measures applied to them.

[DavidGoldwasser]

@macumber FYI - this summarizes what I was talking about. Of course this pre-dates reporting measures. Reporting measures could come from PAT fixed measures.

If we go the OSM approach then I guess fixed E+ measures from PAT would have to be supported, for things like tariffs and QAQC. (I think, but am not sure that this is how it works now. Although currently you may also get fixed PAT OSM measures?)

[Matt32258]

Hello, I am not sure if this is what has happened but I was using the always run Reporting Measure = "Calibration Reports" successfully while in OS. I ran PAT and did not get the reports and graphs I was expecting. I did go back to the Sketchup GUI and modify geometry (eliminate some degenerate surfaces) and now OS does not give me End Use Breakdown and Calibration Report only shows existing. Does this mean my model is now corrupted? Are there lines that have been added to the .idf that need to be deleted? Thank you, Matt

[macumber]

@Matt32258 If the measures in your OSM become corrupt you can delete the companion directory and then open your OSM file. This will create a new companion directory and you can re-add the measures. We are going to try to fix #754 for the 1.6.0 release

[macumber]

@DavidGoldwasser @asparke2 is this still a valid issue or did Elaine fix it?

[asparke2]

I don't think that it has been changed. Looking at it now, I think we need to discuss before making any changes, since people have workflows that take the current behavior into account.

[MatthewSteen]

@macumber @Matt32258 - as of 1.5.0 I haven't been able to permanently disassociate measures from an OSM by simply deleting the directory. They will be removed from the OSM file measure tab but will reappear in the newly created directory when the OSM is reopened.

If you instead copy/paste the OSM to a new location the measures will be permanently disassociated from the model and the scripts folder will be empty. Without doing this the scripts folder will contain a "history" of the measures added/deleted/added... by the user.

[macumber]

@MatthewSteen that is very odd, there is no information about the measures stored in the OSM file, it is all in that directory. Can you try this again? You may have to close the OpenStudio app before deleting the directory and then restart the OpenStudio app to re-open the OSM.

[MatthewSteen]

@macumber - I couldn't reproduce in 1.5.0, another phantom issue I guess, will try to document better in the future.

[Matt32258]

Hi Folks, I appreciate the advise. I created a new blank folder copied just the .osm before running again. The OS results summary still has the drop down for Calibration Report but does not show current data from the simulation. I am also not allowed to add the BCL Calibration Reports back to the model because it "conflicts with the built in measure". I am going to try again and shut my machine down. I realize that temporary folders are used extensively and I wonder if this might be affecting program control. Best, Matt

[Matt32258]

Greetings all,
The .osm seems to be hard wired to run a calibration report ("built in") since it was used once before with the model. However, it does not show the model results. I have uninstalled and re-installed OS and deleted all files related to Measures and the Measures folders. I am going to send the file to another computer that has never run PAT or Measures or OS 1.5. and see if I can shake this bug. Please send me information on how control is coded in the .osm and maybe I can re-set this to begin cleanly. Thank you, Matt

[Matt32258]

I mean to say OS results are displayed but the Calibration Report does not. Also interesting to note but when I was seeing the reports it would never pick up the month of February. I will let you know how the model runs on anther computer. Thank you, Matt

[Matt32258]

Follow up. OS does retain the measure Calibrate. It will not operate or let you re-set once attached or inserted to the .osm file.

[Matt32258]

Problem solved. I believe some changes or maybe a save operation left the required year input blank. But this raises the question of having data that might start in the middle of the year. Thank you, Matt